Three adaptive discrete least squares cubic spline procedures for the compression of data

Abstract Three algorithms are developed for the compression of one-dimensional data by loosely coupled cubic splines. spline knots are located by adaptive iterative searches. The methods are applied to the compression of profile retrieval coefficients obtained from NOAA radiosonde/radiometric data. Additional compression is achieved by error acceptable, significant figure rounding, coupled with storage in integer strings. The results are compared with compression by finite-length discrete Fourier series computed by FFT procedures.